Application and Verification of ECMWF Products 2018
Total Page:16
File Type:pdf, Size:1020Kb
UK UK Application and verification of ECMWF products 2018 Met Office. Contributing authors: Dan Suri, Marion Mittermaier, Rob Neal, Ric Crocker, Helen Titley and Becky Stretton. 1. Summary of major highlights ECMWF output is widely used by the Met Office at many timescales, especially the week 2 period where the Met Office is particularly dependant on ECMWF output. The heaviest users of ECMWF output are the Chief and Deputy Chief Meteorologists, who provide forecast guidance across the organisation for areas around the world and hydrometeorologists at the Flood Forecasting Centre. ECMWF output also forms part of 'BestData', a multi-model blend producing site-specific forecasts available, for example, on the Met Office public website. Meanwhile, the Met Office have been actively engaged in producing some novel and innovative tools to help operational meteorologists visualize ensemble output, creating and verifying regional weather regime forecasts via a tool known internally as the Decider system and developing a global hazard map, which includes a focus on tropical cyclones. 2. Use and application of products 2.1 Post-processing of ECMWF model output 2.1.1 Statistical adaptation Nothing to report. 2.1.2 Physical adaptation Nothing to report. 2.1.3 Derived fields The end of February and first days of March 2018 saw the most severe outbreak of cold winter weather in the UK for several years, as well as over much of western Europe, compounded by Storm Emma which came from the south and generated exceptional blizzards and freezing rain in the south of the country. The cold outbreak had been well predicted in advance, firstly from long-range predictions of a Sudden Stratospheric Warming increasing the risk of blocking, and then two weeks ahead from the Decider system. Decider clusters ECMWF and other global ensembles according to weather patterns (Neal et. al., 2016), and identified very high probabilities of cold easterly flows up to two weeks ahead (Fig 1). In fact the real cold air came a few days later than first anticipated but by the end of the week before it was very clear that a major winter outbreak was coming. By a 7 day lead time the Met Office Global Hazard Map (Fig 2; Robbins and Titley, 2018), which post-processes ECMWF data, had very high probabilities of cold wave conditions across large parts of Europe. Forecasting throughout the week exploited a combination of ECMWF and Met Office models, and a few others including NCEP, to provide multi-model ensemble guidance at all time ranges. 1 UK UK Fig.1 Probabilistic weather pattern forecast from the 0000Z ECMWF medium range run on 13th February 2018, showing high confidence forecast transition to weather pattern in week 2 (left image), which can be described as an anticyclonic easterly pattern with colder-than-average temperatures (top right image). Forecast confidence was unusually high for week 2 (bottom right image). Fig.2 Probability of cold wave conditions at a 7 day lead time as shown by the Met Office Global Hazard Map based on the ECMWF 0000Z medium range run on 20th February 2018. Probabilities show cold wave conditions widespread over much of Europe. 2 UK UK 2.2 ECMWF products 2.2.1 Use of Products At the Met Office, ECMWF output is used alongside the Met Office's own model suite. This suite includes an hourly-cycling, convection-permitting model (UKV), a 12 member convection-permitting ensemble (MOGREPSUK), the global model, run 4 times a day with horizontal resolution of approximately 10 km and an 18 member ensemble, MOGREPS-G. Visualization of ECMWF output by Met Office operational meteorologists is via a number of mechanisms, the primary vehicles being ecCharts and ingestion of a subset of ECMWF HRES into the Visual Weather forecaster workstations. Visualization is supplemented by use of ECMWF’s main products web page and some internal web- based tools and diagnostics, for example classifying ensemble members by synoptic type to objectively assess signals (this product is known as Decider), particularly for severe weather (Neal et. al. 2016) and for coastal flood risk (Price 2017), driving a number of downstream forecasting aids for fluvial, pluvial and tidal flooding and use of multi-model ensembles in, for example, tropical cyclone forecasting (Titley and Neal 2017). Meanwhile, ECMWF model data is blended with Met Office NWP to produce what is known in-house as 'BestData', a database of site- specific forecasts for thousands of sites used for many applications including, for example, forecasts available on the Met Office public web site. In BestData, ECMWF data is the primary source of data for forecasts for week 2. Regarding Met Office forecasting operations, over-arching, authoritative guidance for the shorter term (Day 2) is provided by the Chief Operational Meteorologist, who has ultimate accountability for forecast output and is responsible for operational delivery of National Severe Weather Warning Service (NSWWS) impacts-based warnings. The Chief Operational Meteorologist is supported, on shift, by two Deputy Chiefs. One focuses on UK weather on time scales from Day 2 onwards through medium range and into monthly and seasonal time scales. The other focuses on providing guidance on global forecast matters across multiple time-scales, operating in what is known as the Global Guidance Unit; here customers include Met Office meteorologists working away from the UK, UK government departments, to whom support is provided during periods of severe weather such as the 2017 Atlantic Hurricane season, and cross-European projects such as ARISTOTLE. Other forecast teams, for example aviation, media and marine, then use this guidance to influence their own more customer-specific output. These forecast teams are spread across the UK with some operational meteorologists based at UK military bases, some embedded with clients, for example at some UK airports and transport agencies. some at the main Operations Centre in Exeter or in the Aberdeen office. In addition, there are a number of operational teams at locations around the world (e.g. Falkland Islands). The Flood Forecasting Centre, meanwhile, is a joint partnership with the Environment Agency set up in the wake of severe flooding over parts of the UK in 2007 and comprises of a team of dedicated hydrometeorologists providing advice on fluvial, pluvial and tidal flooding. EFI and meteograms plotted with M-Climate are very popular with the Global Guidance Unit who require an efficient means of visualization severe weather potential around the world and associated climatic context. Cyclone database products (Hewson and Titley 2010) remain an effective means of visualizing extratropical cyclone details. Objective fronts and Dalmatian plots showing low pressure centres and intensity are two of the most popular elements used, especially in medium range forecasting. Meanwhile, diagnostics mentioned above classifying ensemble members by synoptic weather type are immensely useful in decision-making, allowing the operational meteorologist to quickly draw out the main themes and trends shown in ensemble output. Field modification software (Carroll 1997) is used by the Chief and Deputy Chief Operational Meteorologists to produce a graphical best estimate of weather conditions over the UK, a process which draws on a number of different elements of NWP, including ECMWF. Severe weather will be accompanied by the issue of National Severe Weather Warnings Service Warnings (NSWWS). This impacts-based warnings service is aimed at both the general public and resilience/emergency services and makes use of a colour-coded 4x4 matrix to communicate a sense of both likelihood and severity of weather conditions. The range of elements for which warnings are issued has been extended in summer 2018 from the original five elements (namely wind, rain, snow, ice and fog) to also include thunderstorms and lightning. Meanwhile, the capability to issue warnings from five to seven day lead times has also been recently introduced. ECMWF data is used alongside Met Office data in the consideration of NSWWS, and a web-based tool using ensemble output, known as EPS-W (Neal et. al. 2014), can be used to generate first-guess warnings. On example of where ECMWF output proved particularly useful was during the severe cold spell in late February and early March 2018. Use of some ECMWF output during this period has been detailed in Section 2.1.3. Meanwhile, preceding the arrival of Storm Emma into the UK on 28th February 2018 were high confidence forecasts for significant snowfall across many areas. Also, unusually for the UK, was a significant forecast signal for freezing rain to affect parts of Southern England and Wales. The freezing rain risk was highlighted well by the ECMWF precipitation type bar plots (Fig 3), which were used by Met Office Chief and Deputy Chief Operational Meteorologists, this new diagnostic lending confidence to signals for freezing rain present in Met Office NWP. 3 UK UK Fig.3 ECMWF 1200Z run on 27th February 2018: Probability of precipitation type for Fishguard (left) and London (right) highlighting the significant possibility of freezing rain or ice pellets on Thursday night across parts of the SW. 2.2.2 Product requests Feedback from the Met Office forecasting community is dominated by requests for faster performance of ecCharts. There has also been positive feedback regarding the recent introduction of outputs such as precipitation type diagnostics, lightning diagnostics, vertical profiles in ecCharts and IVT outputs. One of the main foci for product development at the Met Office using ECMWF output is the use of multi-model ensembles and tools to identify impacts and significant work is being undertaken to develop a multi-model blended probabilistic post-processing system. Any products aiding this effort would clearly be of use. Another area where input from ECMWF is sought is in developing tools and products to monitor ongoing and developing hazards, improved or novel methods to get better statistics of extremes to contextualise weather events and further developments to reanalysis and/or reforecasts for climate context and calibration.